i7-8086K

[kbk_75]

So I just delided my 8086K and applied Conductonaut / Kryonaut. Managed to go from a stable 5.2GHz all core, no AVX offset @ 1.38V, 5.2 uncore to 5.4GHz, no AVX offset @ 1.50V, 4.7 uncore. Temps are going up to 91-93C while stress testing so I won’t be using this setting 24x7 for sure, just wanted to see how far I could push this chip on air. It can also do a stable 5.3 GHz all core, no AVX offset @ 1.45V, 5.2 uncore. That’s a little high for 24x7 use as well, so I might drop it to 5.3GHz all core, no AVX offset, 1.38V @ 4.7 or so uncore (haven’t tested 4.8 or 4.9 uncore yet).

I’m pretty happy with this CPU, I seem to have got a decent one! Will play around with memory speed and timings tomorrow along with bus frequencies to see how much I can get outta it.

 

[hat]

Wow, second post in over 10 years...

Wouldn't 1.38v even be a bit high for this chip?

 

[cadaveca]

Wow, second post in over 10 years...

Wouldn't 1.38v even be a bit high for this chip?

That's why they use that silly paste... it really does prevent overvolting. He mentioned running 1.5V... That's hilarious. But the temps.. now that's interesting though. The temp drop allows for so much more voltage, but those gains aren't really worth all that added voltage and heat, really, are they?

Shows how well Intel is binning these chips though, and for such a special CPU, that's really important. That chip, delidded, is gonna run super cool @ 5.0 GHz on all cores, and that's quite nice.

 

[xorbe]

5.2 at 1.38v is really good. Turns out my second cpu can't hold 5.2 no matter how much voltage. Well it ran CB in windows, but can't compile source code in Linux. So I still got a bottom 40% sample, eh.

 

[R-T-B]

That's why they use that silly paste... it really does prevent overvolting. He mentioned running 1.5V... That's hilarious.

Not according to Intel. It's actually "in spec":

PS: And then for comparison you have my coffee lake 8700k that can't do 4.8Ghz all cores in summer without at least 1.32+ volts... lol.

 

[kbk_75]

Wow, second post in over 10 years...

Wouldn't 1.38v even be a bit high for this chip?

Haha yup, guess I never posted back when I built my 2600K rig in 2011! That chip has run flawlessly at 4.6 GHz for 7 1/2 years at 1.42V. Different animal, I know, but I think 1.38V for everyday use shouldn’t degrade the IMC on this 8086K. Thoughts?

I do plan on keeping this chip for a while unless something else comes out that absolutely destroys it in the gaming arena, which I, honestly, cannot see happening anytime soon.

That's why they use that silly paste... it really does prevent overvolting. He mentioned running 1.5V... That's hilarious. But the temps.. now that's interesting though. The temp drop allows for so much more voltage, but those gains aren't really worth all that added voltage and heat, really, are they?

Shows how well Intel is binning these chips though, and for such a special CPU, that's really important. That chip, delidded, is gonna run super cool @ 5.0 GHz on all cores, and that's quite nice.

I’d be a cynic and say it has nothing to do with anything other than saving a buck on each chip they sell.

I got a 15C drop after my delid Conductonaut / Kryonaut vs before with AS5. I’m running an NH-D15 so that gives you some perspective for the temps, plus it’s around 25-26C ambient right now and I’m still testing the whole rig outside my HAF-X case. Might get a little hotter once it’s all assembled inside the case and/or it gets warmer here, which it will in a couple months.

5.2 at 1.38v is really good. Turns out my second cpu can't hold 5.2 no matter how much voltage. Well it ran CB in windows, but can't compile source code in Linux. So I still got a bottom 40% sample, eh.

Is that 5.2 with no AVX offset or with an offset? I am seeing a ~.05V drop for each AVX offset I try. Might try to push for 5.5GHz today with an offset of -2 just to see if this thing will manage it! =)

Not according to Intel. It's actually "in spec":

View attachment 104000

PS: And then for comparison you have my coffee lake 8700k that can't do 4.8Ghz all cores in summer without at least 1.32+ volts... lol.

From what I’ve read people have been getting 8086K units with production dates in late 2017, so Intel has been planning this chip for a while and binning the best CPUs as 8086Ks. I contemplated getting an 8700K and hoping for some luck but then went the 8086K route and I’m glad I did. I do not get all the hate this CPU is getting, tbh. They are cherry picked 8700Ks and if I have managed to get to 5.4GHz stable with AVX loads it proves that they are unquestionably the cream of the 8700K crop. May not be worth paying extra for, for most, but nobody’s forcing you to buy the top spec desktop part, right?

This chip ran 5.0GHz stable with no AVX offset but 4.5 uncore at around 1.21V 5 mins after I first fired it up, so I guess it is a good one.

 

[cadaveca]

Not according to Intel. It's actually "in spec":

View attachment 104000

PS: And then for comparison you have my coffee lake 8700k that can't do 4.8Ghz all cores in summer without at least 1.32+ volts... lol.

You are correct sir, but it's not as simple as just upping to voltage, and everything is good. Current limits need adjusting as well, and this does not happen when you overclock. It does should a CPU have this VID assigned, but a VID and actual voltage supplied to a CPU do not always coincide. Thing like current limits and loadline calibrations are big factors in this. I had those whitepapers with such info before the launch of Z370. That's the sort of stuff signing NDAs gets you.

But yeah, clearly these CPUs seem on average much better than 8700K, and very much reminds me of the Devil's Canyon 4790K CPUs, and how 4790K was binned for 4.6 GHz, which is a fair bit higher than the 4770K was, as well as being higher than their stock clocks. Intel says they binned 8086K CPUs for over 5 GHz, but how over was not disclosed. Could be simply 5 GHz within a target voltage that leaves some overhead, which could leave some CPUs barely able to get over 5 GHz (which I have seen some reports of, although they are few are far between).

 

[FireFox]

5.4GHz, no AVX offset @ 1.50V,

I will give it a shot at 1.50V for 5.5GHz

Of course not Delidded

Not according to Intel. It's actually "in spec"

Link please.

 

[R-T-B]

I mean, Dave, I would think you'd know me well enough to know I also know how to calculate a wattage. Yes, current is relevant, yes VIDs consider this and static voltage does not. These are kind of "duh" things for me frankly, no offense intended. Assuming you mean them for a more general reader then.

I was just kind of giggling at how you exclaimed "1.5 volts!" as if that meant everything. It was worthy of a small jab even if we both know you know better.

Link please.

I have it as a pdf on my phone. It's an official intel doc though, somewhere. Can't google it up tonight (again, on my phone) but will try tomorrow.

EDIT: Here is the complete, unabridged doc. Aparently googling on my phone is easier than I remember.

Page 110, specifically.


https://www.intel.com/content/www/u...cessor-family-s-platform-datasheet-vol-1.html

 

[FireFox]

https://www.intel.com/content/www/u...cessor-family-s-platform-datasheet-vol-1.html

Thanks.

Later on turning on the Chiller and increasing voltages

 

[R-T-B]

Thanks.

Later on turning on the Chiller and increasing voltages

Keep in mind as Dave pointed out, that is for stock products using vids with current limits in place.

Yes you can go higher than many think, especially with a chiller. I would not exceed 1.4v for 247 though no matter the cooling.

 

[FireFox]

for 247

Joke.

I would not exceed 1.4v

It's just for test.

 

[kbk_75]

I will give it a shot at 1.50V for 5.5GHz

Of course not Delidded

What cooler are you running? I could not even contemplate 1.45V + before I delided my chip. That’s a real golden sample you have if it can take 1.5V without a delid and not next to instantly hit max temp!

 

[dj-electric]

Just got my hands on my ES 8086K unit. Its a little unfortunate to me that its an ES, so it holds much less value than a retail box that celebrate it, but i hope to get good numbers. Will test it without any modding and a D15 and report my findings - i am less of a ride or die runs and more of a "can we get this to 5Ghz on air and have decent system noise compered to 8700K".

 

[FireFox]

What cooler are you running?

Waterchiller.

 

[kbk_75]

i am less of a ride or die runs and more of a "can we get this to 5Ghz on air and have decent system noise compered to 8700K".

You should get well above 5GHz on a D15 with virtually zero system noise. My PSU and GPU fans don’t even spin when I test the CPU and the D15 is damn near silent sitting one foot from me on my desk! Good luck with your tests!

 

[FireFox]

Just got my hands on my ES 8086K unit. Its a little unfortunate to me that its an ES, so it holds much less value than a retail box that celebrate it, but i hope to get good numbers. Will test it without any modding and a D15 and report my findings - i am less of a ride or die runs and more of a "can we get this to 5Ghz on air and have decent system noise compered to 8700K".

To hit 5.0GHz/5.1GHz is pretty easy but i don't know if it is the same thing for an ES.

 

[dj-electric]

I mean, there's not much of an engineering to be done. This isn't some early hardware of any kind, not since the 8700K exists in engineering almost a whole year earlier. I think its marked as ES to be used internally or by media instead of being sold outside.

 

[xorbe]

Is that 5.2 with no AVX offset or with an offset? I am seeing a ~.05V drop for each AVX offset I try. Might try to push for 5.5GHz today with an offset of -2 just to see if this thing will manage it! =) [...] This chip ran 5.0GHz stable with no AVX offset but 4.5 uncore at around 1.21V 5 mins after I first fired it up, so I guess it is a good one.

No AVX offset, as Linux seems to keep the offset active 100% of the time, so it becomes pointless. 1.21 @ 5.0 is incredible, that's 0.08v less than mine, that's a 14% power savings which makes it a lot easier to keep temps down for you.

 

[FireFox]

I wanted to run a test but the Dew point is 11c

 

[dj-electric]

Time to isolate, dude

 

[FireFox]

Time to isolate, dude

Nah, i will wait till the Dew point will be lower, when i said the Dew point is 11c i meant in the Room.

This is a Platinum CPU

i7 8086K 5.5 ghz / 5500 Mhz running Cinebench R15 on 1,408v

 

[xorbe]

YOU HAVE TO DELID!!!

Slightly sanded under-edge of cap to make sure it would rest directly on die. Slammed back on with NT-H1 and nothing else to hold the cap (ghetto, eh). It really is truly smashed against the die, lol. Wowzers, it can run Prime95 AVX small FFT @ 5.0 without it immediately doing 100C within 3 seconds, niiiiiiice! It ran AVX small FFTs 1 minute each for 15 minutes, bounces 82-84C. Oh man so nice. If I am not mistaken it runs avx slightly cooler than non-avx before delid.

 

[FireFox]

YOU HAVE TO DELID!!!

Why should I?

 

[xorbe]

Why should I?

Still using play-doh paste?